Comparison of Relative Quantification of Monoclonal ntibody N-glycans Using Fluorescence and MS Detection pplication Note iotherapeutics & iologics uthors scar Potter and Gregory Staples gilent Technologies, Inc. Santa Clara, California, US bstract This application note describes the analysis of monoclonal antibody N-glycosylation using a novel instant mass tag (InstantPC) that permits detection using both fl uorescence and mass spectrometry (MS). We have optimized a HILIC separation method for the purpose of comparing the two detection methods. ur results indicate that fl uorescence and MS relative quantifi cation of InstantPC-labeled glycans is highly similar. The sample prep procedure was conducted in quadruplicate for two different samples, and the results showed outstanding reproducibility with low RSDs even for minor components. The high MS sensitivity afforded by the InstantPC label facilitates identifi cation of unknown glycans using accurate mass and tandem MS.
Introduction Monoclonal antibodies (mbs) are modified by N-glycans during biosynthesis in cell culture. Typical mbs contain two N-glycosylation sites, one in each of the Fc regions of the molecule. Some mbs contain additional glycosylation sites, including N- or -glycosylation in the Fab region. Glycans can affect the function of the mb, so it is important to monitor the glycosylation profile using appropriate analytical methods. Popular methods for glycan analysis involve NMR, CE-LIF, HPLC with fluorescence detection (FLD), and more recently, LC/MS. oth CE-LIF and HPLC-FLD require that the glycans are labeled with a dye to permit optical detection. Conventionally, the dyes that have been used also increase the ionization efficiency of glycans in comparison to the unlabeled species, but only to the point where the most abundant compositions can be detected using MS. More recently, a novel dye (InstantPC from Prozyme Inc., depicted in Figure 1) has been developed, which moderately improves fluorescence activity and greatly improves ionization efficiency for MS analysis. Using such a tag, researchers can now use MS (in the form of accurate mass or tandem MS) for identification of glycans from LC separations. Furthermore, they have the option of relative quantification using MS rather than fluorescence detection. InstantPC Labeled mb N-glycans gilent 129 LC with Fluorescence Detection gilent dvanceio Glycan Mapping Column N Figure 1. N H N Diagram of InstantPC (ProZyme, Inc), an amine reactive instant label for fluorescence and MS detection of glycans. gilent 655 ifunnel Q-TF LC/MS This application note investigates the performance of InstantPC in the context of relative quantification of N-glycans released from two mb preparations. Quadruplicate samples of the mb samples were processed using the InstantPC kit from ProZyme, Inc. The samples were then separated by HILIC on a UHPLC system using FLD and MS detection. The LC separation conditions were optimized for maximum chromatographic separation. In doing so, the goal was to decrease the number of overlapping peaks that would otherwise not be discernable using FLD detection alone. s a result, we were able to compare the relative quantification results from the two detection methods for nearly all significant glycan structures. ccurate mass and tandem MS spectra were acquired for all glycan compositions, and were used for identification of the glycans present in the mb preparations. Figure 2 shows the entire workflow. Figure 2. gilent Mass Profiler Software Workflow used for identification and quantification of InstantPC-labeled N-glycans from mbs. 2
Experimental The gilent LC/MS System used in this work comprised the following modules: gilent 129 Infinity inary Pump (G422) gilent 129 Infinity High Performance utosampler (G4226) with an gilent 129 Infinity Thermostat (G133) gilent 129 Infinity Thermostatted Column Compartment (G1316C) gilent 126 Infinity Fluorescence Detector (G1321) MS system gilent 655 ifunnel Q-TF LC/MS system with dual-nebulizer JS source Columns gilent dvanceio Glycan Mapping column, 2.1 15 mm, 1.8 µm connected to a second dvanceio Glycan Mapping column, 2.1 1 mm Software gilent PCDL Manager (Version.7. uild 724.) and gilent Mass Profiler (Version.7.1 uild 99.) gilent MassHunter Workstation Software, Version.5.1, uild 5.1.5125.1 Solvents and samples ll reagents and solvents used were of the highest purity available. Chromatographic conditions Parameter Value Mobile phase 5 mm ammonium formate ph 4.4 Mobile phase cetonitrile Gradient Time (min) % 75 32 69 48 6 48.5 25 5.5 25 52 75 utosampler temperature 4 C FLD Ex. 285 Em. 345 Injection 2 µl in 2 % DMS (1 µg of IgG equivalent) Column temperature 4 C Flow rate.4 ml/min 3
Results and Discussion FLD chromatograms from both mb 1 and mb 2 revealed that each molecule was modified by a very similar set of glycoforms, as shown in Figure 3. Some structures have been annotated in the figures, and represented by symbols according to the guidelines of the Consortium for Functional Glycomics (CFG) [1]. Using accurate mass and tandem mass spectrometry information, FLD peaks were assigned to glycan compositions in the form: HxNxFxSgx + Core H = galactose or mannose, N = N-acetylglucosamine, F = fucose, Sg = N-glycolylneuraminic acid, and Core = trimannosyl core common to all N-glycans. 3 25 Zoom 2 FLD Intensity 15 1 1 15 2 25 3 35 5 1 15 2 25 3 35 Time (min) 2 15 Zoom FLD Intensity 1 1 15 2 25 3 35 5 1 15 2 25 3 35 Time (min) Figure 3. FLD chromatograms of InstantPC-labeled N-glycans released from mb 1 and mb 2. ) FLD chromatogram for mb 1. ) FLD chromatogram for mb 2. 4
ased on peak area from the FLD chromatograms, each composition was quantified as a relative sum percentage based on the total FLD area for all compositions. The results are shown as a histogram in Figure 4. and reasonable resolution from neighboring peaks. ne exception was the pair H2N1F1Sg1 + Core/H2N3F1 + Core. These coeluting compositions were abundant enough to merit inclusion in the FLD quantification. Therefore, the FLD signal area from this peak was divided into two portions according to the relative abundance of each as determined by MS. verall, 21 glycan compositions were quantified based on the FLD data. The criteria used for inclusion in the FLD quantification were a relative abundance of.1 % or greater, 5 4 1 6 3 2 1 45 4 35 3 25 1 2 15 1 5 Figure 4. ) Relative FLD quantification of mb 1 glycans. ) Relative FLD quantification of mb 2 glycans. Error bars represent ± standard deviation of quadruplicates having gone through the entire workflow. Integration of FLD signals was performed using gilent MassHunter Qualitative nalysis Software. Insets show the same data zoomed to better display components with <1 % relative abundance. 5
InstantPC imparts high ionization efficiency to N-glycans. Thus, it is possible to perform relative quantification using the peak area from extracted ion chromatograms from MS detection. To assess this possibility, we compared FLD chromatograms with ion chromatograms. Figure 5 shows that the FLD and MS chromatograms were highly similar. There is a corresponding MS peak for every FLD peak that was detected. Encouraged by the high similarity seen in Figure 5, we performed relative quantification of glycans from mb 1 and mb 2 based on the MS data. Ion chromatograms for each feature (defined as a mass-retention time pair, which includes signals from all charge states and adducts) were created using gilent Mass Profiler software. In this case, no lower threshold for detection was imposed. The features determined using Mass Profiler were identified using a Personal Compound Database (PCD) constructed for these experiments. FLD FLD Zoom 9 14 19 24 29 34 39 9 14 19 24 29 34 39 C MS D MS Zoom 9 14 19 24 29 34 39 9 14 19 24 29 34 39 Figure 5. Comparison of FLD and MS chromatograms for mb 1. ) FLD chromatogram of mb 1 glycans. ) Zoom of FLD chromatogram of mb 1 glycans. C) MS chromatogram of mb 1 glycans. D) Zoom of MS chromatogram of mb 1 glycans. 6
The PCD contains accurate mass and retention time information for mb glycans. The database was constructed based on a combination of tandem MS information from the current work in addition to knowledge of glycan biosynthetic rules. Figure 6 shows an example of the utility of tandem MS for assigning glycan compositions. In particular, the example shown in Figure 6 illustrates a common case where mass alone may be insufficient for assignment of composition, due to the fact that the mass of NeuGc + fucose is isobaric with that of Neuc + galactose. Tandem MS resolves the ambiguity, because the presence of the fragment ion at m/z 673 provides strong evidence that the structure contains an antenna with NeuGc. Intensity (counts) Compound 14 precursor 1 2 3 4 5 6 7 8 9 m/z 1, 1,1 1,2 1,3 1,4 1,5 1,6 1,7 Intensity (counts) Compound 47 precursor 1 2 3 4 5 6 7 8 9 m/z 1, 1,1 1,2 1,3 1,4 1,5 1,6 1,7 Figure 6. Tandem MS data were acquired for all glycans. MS/MS aided in compound identification when accurate mass was insufficient. The two examples above are consistent with gal-gal and outer arm fucose () and NeuGC () modifications. 7
Figure 7 shows the results of the MS-based quantification of mb 1 and mb 2 glycans. s a result of the mass selectivity provided by Q-TF detection, it was possible to quantify more compositions than from the FLD detection. In this case, a total of 35 compositions were quantified. The average RSD was 3.2 % for mb 1 and 3.9 % for mb 2 for all features independent of abundance. For those features equal to or greater than.1 % relative abundance, RSDs were 2.7 % and 3.4 % respectively. 6 5 4 3 2 1 1 N2F1 + Core 1 H2N3F1 + Core 6 5 4 3 2 1 1 N2F1 + Core 1 H2N3F1 + Core Figure 7. 11 H1N2F2 + Core 12 H2N2F2 + Core 13 H3N2F1 + Core 14 H3N2F2 + Core 15 H3N3F1 + Core 11 H1N2F2 + Core 12 H2N2F2 + Core 13 H3N2F1 + Core 14 H3N2F2 + Core 16 H4N2F1 + Core 17 H4N3F1 + Core 18 N2 + Core 2 H4N2F1 + Core 21 N1 + Core 22 H1N1 + Core 35 H1N1F1Sg1 + Core 36 H1N2F1Sg1 + Core 38 H2N2F1Sg1 + Core 15 H3N3F1 + Core 16 H4N2F1 + Core 17 H4N3F1 + Core 18 N2 + Core 2 H4N2F1 + Core 21 N1 + Core 22 H1N1 + Core 35 H1N1F1Sg1 + Core 36 H1N2F1Sg1 + Core 38 H2N2F1Sg1 + Core 1 1 1 N2F1 + Core 1 H2N3F1 + Core 11 H1N2F2 + Core 12 H2N2F2 + Core 13 H3N2F1 + Core 14 H3N2F2 + Core 15 H3N3F1 + Core 3a H2N1F1 + Core, 3a 16 H4N2F1 + Core 17 H4N3F1 + Core 3b H2N1F1 + Core, 3b 18 N2 + Core 2 H4N2F1 + Core 4 N1F1 + Core 21 N1 + Core 22 H1N1 + Core 4 H3N2F1Sg1 + Core 35 H1N1F1Sg1 + Core 45 Man5 36 H1N2F1Sg1 + Core 38 H2N2F1Sg1 + Core 47 H2N1F1Sg1 + Core 3a H2N1F1 + Core, 3a 3b H2N1F1 + Core, 3b 48 H3N1 + Core 4 N1F1 + Core 4 H3N2F1Sg1 + Core 5 H1N1F1 + Core 45 Man5 51 H4N1 + Core 47 H2N1F1Sg1 + Core 1 N2F1 + Core 48 H3N1 + Core 1 H2N3F1 + Core 52 H4N1F1 + Core 5 H1N1F1 + Core 11 H1N2F2 + Core 51 H4N1 + Core 53 Man3 12 H2N2F2 + Core 52 H4N1F1 + Core 13 H3N2F1 + Core 53 Man3 54 Man3F1 14 H3N2F2 + Core 54 Man3F1 15 H3N3F1 + Core 55 H1N1F2 + Core 55 H1N1F2 + Core 3a H2N1F1 + Core, 3a 16 H4N2F1 + Core 56 H6N3F1 + Core 17 H4N3F1 + Core 56 H6N3F1 + Core 6 H2N1F1 + Core 3b H2N1F1 + Core, 3b 18 N2 + Core 8 N3F1 + Core 6 H2N1F1 + Core 2 H4N2F1 + Core 9 H1N3F1 + Core 4 N1F1 + Core 21 N1 + Core 19 H1N2 + Core 8 N3F1 + Core 23 H2N1 + Core 4 H3N2F1Sg1 + Core 22 H1N1 + Core 35 H1N1F1Sg1 + Core 9 H1N3F1 + Core 49 H3N1F1 + Core 45 Man5 36 H1N2F1Sg1 + Core 38 H2N2F1Sg1 + Core 19 H1N2 + Core 47 H2N1F1Sg1 + Core 3a H2N1F1 + Core, 3a 3b H2N1F1 + Core, 3b 23 H2N1 + Core 48 H3N1 + Core 4 N1F1 + Core 49 H3N1F1 + Core 5 H1N1F1 + Core 4 H3N2F1Sg1 + Core 45 Man5 51 H4N1 + Core 47 H2N1F1Sg1 + Core 48 H3N1 + Core 52 H4N1F1 + Core 5 H1N1F1 + Core 51 H4N1 + Core 53 Man3 52 H4N1F1 + Core 53 Man3 54 Man3F1 54 Man3F1 55 H1N1F2 + Core 55 H1N1F2 + Core 56 H6N3F1 + Core 56 H6N3F1 + Core 6 H2N1F1 + Core 8 N3F1 + Core 6 H2N1F1 + Core 9 H1N3F1 + Core 19 H1N2 + Core 8 N3F1 + Core 23 H2N1 + Core 9 H1N3F1 + Core 49 H3N1F1 + Core 19 H1N2 + Core 23 H2N1 + Core 49 H3N1F1 + Core Relative MS quantification of InstantPC labeled N-glycans released from mb 1 and mb 2. ) Relative MS quantification of mb 1 glycans. ) Relative MS quantification of mb 2 glycans. Error bars represent ± standard deviation of quadruplicates having gone through the entire workflow. Insets show the same data zoomed to better display components with <1 % relative abundance. 8
Finally, we directly compared the relative quantification of glycans from mb 1 and mb 2 using FLD and MS. Figure 8 shows the results from each method plotted on a single histogram. Conclusion s shown in Figure 8, the relative quantification results from FLD and MS were highly similar. Some small differences in the results from the two methods can be explained by the different numbers of compositions quantified in the two methods (21 from FLD, 35 from MS). ased on the results of this study, the combination of Prozyme s InstantPC label and an gilent LC/MS system provides the researcher with the capability to perform MS-based quantification of glycans from mbs. Still, FLD will likely remain a gold standard detection method for this compound class. In that case, high quality Q-TF MS data greatly facilitate peak assignment by offering accurate mass and tandem mass information for each of the InstantPC-labeled glycans detected using FLD. 6 5 mb 1 1 4 3 2 FLD Quant mb 1 MS Quant mb 2 1 1 2 3 4 5 6 7 8 9 11112131415161718192212223242526272829331323334353637383944142434445464748495515253545556 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3 31 32 33 34 35 36 37 38 39 4 41 42 43 44 45 46 47 48 49 5 51 52 53 54 55 56 6 5 4 mb 2 1 FLD Quant mb 1 MS Quant mb 2 3 2 1 12345678911112131415161718192212223242526272829331323334353637383944142434445464748495515253545556 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3 31 32 33 34 35 36 37 38 39 4 41 42 43 44 45 46 47 48 49 5 51 52 53 54 55 56 Figure 8. Comparison of fluorescence and relative MS abundance (area sum percentage) of InstantPC-labeled N-glycans from mb 1 and mb 2. The X-axis represents individual glycan compositions quantified in the study. 9
cknowledgements We would like to thank NIST for providing the two mb samples used in this work. Reference 1. http://glycomics.scripps.edu/cfgnomenclature.pdf For More Information These data represent typical results. For more information on our products and services, visit our Web site at www.agilent.com www.agilent.com For Research Use nly. Not for use in diagnostic procedures. This information is subject to change without notice. gilent Technologies, Inc., 216 Printed in the US June 29, 216 5991-6958EN